Side-chain Conformations Research Articles

All‐Atom Protein Sequence Design Based on Geometric Deep Learning

AbstractDesigning sequences for specific protein backbones is a key step in creating new functional proteins. Here, we introduce GeoSeqBuilder, a deep learning framework that integrates protein sequence generation with side chain conformation prediction to produce the complete all‐atom structures for designed sequences. GeoSeqBuilder uses spatial geometric features from protein backbones and explicitly includes three‐body interactions of neighboring residues. GeoSeqBuilder achieves native residue type recovery rate of 51.6 %, comparable to ProteinMPNN and other leading methods, while accurately predicting side chain conformations. We first used GeoSeqBuilder to design sequences for thioredoxin and a hallucinated three‐helical bundle protein. All the 15 tested sequences expressed as soluble monomeric proteins with high thermal stability, and the 2 high‐resolution crystal structures solved closely match the designed models. The generated protein sequences exhibit low similarity (minimum 23 %) to the original sequences, with significantly altered hydrophobic cores. We further redesigned the hydrophobic core of glutathione peroxidase 4, and 3 of the 5 designs showed improved enzyme activity. Although further testing is needed, the high experimental success rate in our testing demonstrates that GeoSeqBuilder is a powerful tool for designing novel sequences for predefined protein structures with atomic details. GeoSeqBuilder is available at https://github.com/PKUliujl/GeoSeqBuilder.

All-Atom Protein Sequence Design Based on Geometric Deep Learning.

Designing sequences for specific protein backbones is a key step in creating new functional proteins. Here, we introduce GeoSeqBuilder, a deep learning framework that integrates protein sequence generation with side chain conformation prediction to produce the complete all-atom structures for designed sequences. GeoSeqBuilder uses spatial geometric features from protein backbones and explicitly includes three-body interactions of neighboring residues. GeoSeqBuilder achieves native residue type recovery rate of 51.6 %, comparable to ProteinMPNN and other leading methods, while accurately predicting side chain conformations. We first used GeoSeqBuilder to design sequences for thioredoxin and a hallucinated three-helical bundle protein. All the 15 tested sequences expressed as soluble monomeric proteins with high thermal stability, and the 2 high-resolution crystal structures solved closely match the designed models. The generated protein sequences exhibit low similarity (minimum 23 %) to the original sequences, with significantly altered hydrophobic cores. We further redesigned the hydrophobic core of glutathione peroxidase 4, and 3 of the 5 designs showed improved enzyme activity. Although further testing is needed, the high experimental success rate in our testing demonstrates that GeoSeqBuilder is a powerful tool for designing novel sequences for predefined protein structures with atomic details. GeoSeqBuilder is available at https://github.com/PKUliujl/GeoSeqBuilder.

Integrating 19F Distance Restraints for Accurate Protein Structure Determination by Magic Angle Spinning NMR Spectroscopy.

Traditional protein structure determination by magic angle spinning (MAS) solid-state NMR spectroscopy primarily relies on interatomic distances up to 8 Å, extracted from 13C-, 15N-, and 1H-based dipolar-based correlation experiments. Here, we show that 19F fast (60 kHz) MAS NMR spectroscopy can supply additional, longer distances. Using 4F-Trp,U-13C,15N crystalline Oscillatoria agardhii agglutinin (OAA), we demonstrate that judiciously designed 2D and 3D 19F-based dipolar correlation experiments such as (H)CF, (H)CHF, and FF can yield interatomic distances in the 8-16 Å range. Incorporation of fluorine-based restraints into structure calculation improved the precision of Trp side chain conformations as well as regions in the protein around the fluorine containing residues, with notable improvements observed for residues in proximity to the Trp pairs (W10/W17 and W77/W84) in the carbohydrate-binding loops, which lacked sufficient long-range 13C-13C distance restraints. Our work highlights the use of fluorine and 19F fast MAS NMR spectroscopy as a powerful structural biology tool.

Discovery of Bis-imidazolecarboxamide Derivatives as Novel, Potent, and Selective TNIK Inhibitors for the Treatment of Idiopathic Pulmonary Fibrosis.

Traf2- and Nck-interacting kinase (TNIK) has been identified as a promising therapeutic target for the treatment of fibrosis-driven diseases. Utilizing a structure-based drug design workflow, we developed a series of potent TNIK inhibitors that modulate the conformation of the gatekeeper Met105 side chain and access the TNIK back pocket. The lead optimization efforts culminated in the discovery of the recently reported compound 4 (INS018_055), a novel TNIK inhibitor. This molecule demonstrated excellent activity in both enzymatic and cell-based assays, along with high selectivity in a kinome panel. Further, in vitro and in vivo preclinical studies revealed favorable in vitro and in vivo DMPK properties. Results from multiple cell-based and animal models proved that compound 4 exhibits considerable antifibrotic and anti-inflammatory efficacy. Currently, phase II clinical trials of compound 4 are underway for the treatment of idiopathic pulmonary fibrosis (IPF).

Potent and Protease Resistant Azapeptide Agonists of the GLP‐1 and GIP Receptors

AbstractThe gut‐derived peptide hormones glucagon‐like peptide‐1 (GLP‐1) and glucose‐dependent insulinotropic polypeptide (GIP) play important physiological roles including glucose homeostasis and appetite suppression. Stabilized agonists of the GLP‐1 receptor (GLP‐1R) and dual agonists of GLP‐1R and GIP receptor (GIPR) for the management of type 2 diabetes and obesity have generated widespread enthusiasm and have become blockbuster drugs. These therapeutics are refractory to the action of dipeptidyl peptidase‐4 (DPP4), that catalyzes rapid removal of the two N‐terminal residues of the native peptides, in turn severely diminishing their activity profiles. Here we report that a single atom change from carbon to nitrogen in the backbone of the entire peptide makes them refractory to DPP4 action while still retaining full potency and efficacy at their respective receptors. This was accomplished by use of aza‐amino acids, that are bioisosteric replacements for α‐amino acids that perturb the structural backbone and local side chain conformations. Molecular dynamics simulations reveal that aza‐amino acid can populate the same conformational space that GLP‐1 adopts when bound to the GLP‐1R. The insertion of an aza‐amino acid at the second position from the N‐terminus in semaglutide and in a dual agonist of GLP‐1R and GIPR further demonstrates its capability as a viable alternative to current DPP4 resistance strategies while offering additional structural variation that may influence downstream signaling.

Solvent organization in the ultrahigh-resolution crystal structure of crambin at room temperature.

Ultrahigh-resolution structures provide unprecedented details about protein dynamics, hydrogen bonding and solvent networks. The reported 0.70 Å, room-temperature crystal structure of crambin is the highest-resolution ambient-temperature structure of a protein achieved to date. Sufficient data were collected to enable unrestrained refinement of the protein and associated solvent networks using SHELXL. Dynamic solvent networks resulting from alternative side-chain conformations and shifts in water positions are revealed, demonstrating that polypeptide flexibility and formation of clathrate-type structures at hydrophobic surfaces are the key features endowing crambin crystals with extraordinary diffraction power.

Exciting Novel Polyaspartates: Design, Synthesis, and Photo-Responsive Behavior in Solution and Lyotropic Liquid Crystalline Phase Upon Irradiation with Visible Light.

Many polypeptides form stable, helical secondary structures enabling the formation of lyotropic liquid crystalline (LLC) phases. Contrary to the well-studied polyglutamate, their counterparts based on polyaspartates exhibit a much lower helix inversion barrier. Therefore, the helix sense is not solely dictated by the chirality of the amino acid used, but additionally by the nature and conformation of the polymer sidechain. In this work, polymers responsive to irradiation with visible light are designed achieving conformational transitions from helix-to-coil and helix-to-helix. The synthesis and the application as LLC mesogens of several (co-)polyaspartates bearing ortho-fluorinated azobenzene (FAB) as a photochromic group are presented. Many of the obtained polymers undergo changes in their secondary structure upon E-Z-isomerization of the FAB-containing sidechain. Of special interest are copolymers that exhibit photo-responsive helix inversion without loss of their helical secondary structure. These copolymers form stable LLC phases in helicogenic solvents, where the effect of photo-switching on the macroscopic behavior is studied by NMR spectroscopy. Especially, the irradiation of the different LLC phases of the helix inversion polymers displays a change in the LLC order experienced by the solvent. These peculiar properties are promising for future applications as photo-responsive alignment media for structure elucidation in NMR.

Potent and Protease Resistant Azapeptide Agonists of the GLP-1 and GIP Receptors.

The gut-derived peptide hormones glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) play important physiological roles. Stabilized agonists of the GLP-1 receptor (GLP-1R) and the GIP receptor (GIPR) for the management of diabetes and obesity have generated widespread enthusiasm and have become blockbuster drugs. These therapeutics are refractory to the action of dipeptidyl peptidase-4 (DPP4), that catalyzes rapid removal of the two N-terminal residues of the native peptides, in turn severely diminishing their activity profiles. Here we report that a single atom change from carbon to nitrogen in the backbone of the entire peptide make them refractory to DPP4 action while still retaining full potency and efficacy at their respective receptors. This was accomplished by use of aza-amino acids, that are bioisosteric replacements for a-amino acids that perturb the structural backbone and local side chain conformations. Molecular dynamics simulations reveal that aza-amino acid can populate the same conformational space that GLP-1 adopts when bound to the GLP-1R. The insertion of an aza-amino acid at the second position from the N-terminus in semaglutide and in a dual agonist of GLP-1R and GIPR further demonstrates its capability as a viable alternative to current DPP4 resistance strategies while offering additional structural variety.

Resource analysis of quantum algorithms for coarse-grained protein folding models

Protein folding processes are a vital aspect of molecular biology that is hard to simulate with conventional computers. Quantum algorithms have been proven superior for certain problems and may help tackle this complex life science challenge. We analyze the resource requirements for simulating simplified yet computationally challenging protein folding models on a quantum computer, assessing the feasibility of these existing approaches in the current and near-future technological landscape. We calculate the minimum number of qubits, interactions, and two-qubit gates necessary to build a heuristic quantum algorithm with the specific information of a folding problem. Particularly, we focus on the resources needed to build quantum operations based on the Hamiltonian linked to the protein folding models for a given amino acid count. Such operations are a fundamental component of these quantum algorithms, guiding the evolution of the quantum state for efficient computations. Specifically, we study coarse-grained folding models on the lattice and the fixed backbone side-chain conformation model and assess their compatibility with the constraints of existing quantum hardware given different bit encodings. We conclude that the number of qubits required falls within current technological capabilities. However, the limiting factor is the high number of interactions in the Hamiltonian, resulting in a quantum gate count unavailable today. Published by the American Physical Society 2024

Photoinduced Reversible Disentanglement-Entanglement Transition in partially miscible azobenzene polymer blends

Entanglements, an intrinsic feature of long chains, are critical to the rheological and mechanical properties of polymers. Adding low molecular weight solvents or oligomers into entangled polymers can reduce the entanglements, but the entanglement cannot recover without removing the solvents. Reversible control of the entanglement degree is still a big challenge. In this work, we introduce the azobenzene polymer (Pazo) with reversible photoswitchable side-chain conformation as dynamic diluent into entangled poly(butyl methacrylate) (PBMA) to form a partially miscible blend system. The reduction of entanglement of PBMA by trans-Pazo is similar to a typical solvent, while an unexpectedly stronger tube dilation effect was found for cis-Pazo. Cis-Pazo/PBMA blends exhibit a stronger dependence of plateau modulus on PBMA content, GN(φ)∼φ4.5, which is significantly different from the well-established relation GN(φ)∼φ2 for polymer solutions and blends. The cis-Pazo can disentangle PBMA at a much lower concentration (∼55 %) than the trans-Pazo (>70 %). Pazo content of 55 %–70 % can induce a completely unentangled state of PBMA in Pazo/PBMA blends when Pazo is in the cis state after UV light irradiation, and the entanglement can recover when it is transformed to the trans state. Adding azobenzene polymer does not alter the activation energy of the re-entanglement process.

Conformational disorder in the crystal structure of methyl 2-acetamido-2-deoxy-β-D-glucopyranosyl-(1→4)-2-acetamido-2-deoxy-β-D-glucopyranoside (methyl β-chitobioside) methanol monosolvate.

Methyl 2-acetamido-2-deoxy-β-D-glucopyranosyl-(1→4)-2-acetamido-2-deoxy-β-D-glucopyranoside (methyl β-chitobioside), (IV), crystallizes from aqueous methanol at room temperature to give a structure (C17H30N2O22·CH3OH) containing conformational disorder in the exocyclic hydroxymethyl group of one of its βGlcNAc residues. As observed in other X-ray structures of disaccharides containing β-(1→4) O-glycosidic linkages, inter-residue hydrogen bonding between O3H of the βGlcNAc bearing the OCH3 aglycone and O5 of the adjacent βGlcNAc is observed based on the 2.79 Å internuclear distance between the O atoms. The structure of (IV) was compared to that determined previously for 2-acetamido-2-deoxy-β-D-glucopyranosyl-(1→4)-2-acetamido-2-deoxy-β-D-glucopyranose (β-chitobiose), (III). The O-glycosidic linkage torsion angles, phi (φ) and psi (ψ), in (III) and (IV) differ by 6-8°. The N-acetyl side chain conformation in (III) and (IV) shows some context dependence, with the C1-C2-N-Ccar torsion angle 10-15° smaller for the βGlcNAc residue involved in the internal O-glycosidic linkage. In (IV), conformational disorder is observed in the exocyclic hydroxymethyl (-CH2OH) group in the βGlcNAc residue bearing the OCH3 aglycone, and a fitting of the electron density indicates an approximate 50:50 distribution of the gauche-gauche (gg) and gauche-trans (gt) conformers in the lattice. Similar behavior is not observed in (III), presumably due to the different packing structure in the vicinity of the -CH2OH substituent that affects its ability to hydrogen bond to proximal donors/acceptors. Unlike (IV), a re-examination of the previously reported electron density of (III) revealed conformational disorder in the N-acetyl side chain attached to the reducing-end βGlcNAc residue caused by rotation about the C2-N bond.

Invariant point message passing for protein side chain packing.

Protein side chain packing (PSCP) is a fundamental problem in the field of protein engineering, as high-confidence and low-energy conformations of amino acid side chains are crucial for understanding (and designing) protein folding, protein-protein interactions, and protein-ligand interactions. Traditional PSCP methods (such as the Rosetta Packer) often rely on a library of discrete side chain conformations, or rotamers, and a forcefield to guide the structure to low-energy conformations. Recently, deep learning (DL) based methods (such as DLPacker, AttnPacker, and DiffPack) have demonstrated state-of-the-art predictions and speed in the PSCP task. Building off the success of geometric graph neural networks for protein modeling, we present the Protein Invariant Point Packer (PIPPack) which effectively processes local structural and sequence information to produce realistic, idealized side chain coordinates using -angle distribution predictions and geometry-aware invariant point message passing (IPMP). On a test set of ∼1400 high-quality protein chains, PIPPack is highly competitive with other state-of-the-art PSCP methods in rotamer recovery and per-residue RMSD but is significantly faster.

Data-Efficient Generation of Protein Conformational Ensembles with Backbone-to-Side-Chain Transformers.

Excitement at the prospect of using data-driven generative models to sample configurational ensembles of biomolecular systems stems from the extraordinary success of these models on a diverse set of high-dimensional sampling tasks. Unlike image generation or even the closely related problem of protein structure prediction, there are currently no data sources with sufficient breadth to parametrize generative models for conformational ensembles. To enable discovery, a fundamentally different approach to building generative models is required: models should be able to propose rare, albeit physical, conformations that may not arise in even the largest data sets. Here we introduce a modular strategy to generate conformations based on "backmapping" from a fixed protein backbone that (1) maintains conformational diversity of the side chains and (2) couples the side-chain fluctuations using global information about the protein conformation. Our model combines simple statistical models of side-chain conformations based on rotamer libraries with the now ubiquitous transformer architecture to sample with atomistic accuracy. Together, these ingredients provide a strategy for rapid data acquisition and hence a crucial ingredient for scalable physical simulation with generative neural networks.

Pushed to extremes: distinct effects of high temperature versus pressure on the structure of STEP

Protein function hinges on small shifts of three-dimensional structure. Elevating temperature or pressure may provide experimentally accessible insights into such shifts, but the effects of these distinct perturbations on protein structures have not been compared in atomic detail. To quantitatively explore these two axes, we report the first pair of structures at physiological temperature versus. high pressure for the same protein, STEP (PTPN5). We show that these perturbations have distinct and surprising effects on protein volume, patterns of ordered solvent, and local backbone and side-chain conformations. This includes interactions between key catalytic loops only at physiological temperature, and a distinct conformational ensemble for another active-site loop only at high pressure. Strikingly, in torsional space, physiological temperature shifts STEP toward previously reported active-like states, while high pressure shifts it toward a previously uncharted region. Altogether, our work indicates that temperature and pressure are complementary, powerful, fundamental macromolecular perturbations.

Structure and dynamics of the proton-selective histidine and the gating tryptophan in an inward rectifying hybrid influenza B and A virus M2 proton channel.

The M2 proteins of influenza A and B viruses form acid-activated proton channels that are essential for the virus lifecycle. Proton selectivity is achieved by a transmembrane (TM) histidine whereas gating is achieved by a tryptophan residue. Although this functional apparatus is conserved between AM2 and BM2 channels, AM2 conducts protons exclusively inward whereas BM2 conducts protons in either direction depending on the pH gradient. Previous studies showed that in AM2, mutations of D44 abolished inward rectification of AM2, suggesting that the tryptophan gate is destabilized. To elucidate how charged residues C-terminal to the tryptophan regulates channel gating, here we investigate the structure and dynamics of H19 and W23 in a BM2 mutant, GDR-BM2, in which three BM2 residues are mutated to the corresponding AM2 residues, S16G, G26D and H27R. Whole-cell electrophysiological data show that GDR-BM2 conducts protons with inward rectification, identical to wild-type (WT) AM2 but different from WT-BM2. Solid-state NMR 15N and 13C spectra of H19 indicate that the mutant BM2 channel contains higher populations of cationic histidine and neutral τ tautomers compared to WT-BM2 at acidic pH. Moreover, 19F NMR spectra of 5-19F-labeled W23 resolve three peaks at acidic pH, suggesting three tryptophan sidechain conformations. Comparison of these spectra with the tryptophan spectra of other M2 peptides suggests that these indole sidechain conformations arise from interactions with the C-terminal charged residues and with the N-terminal cationic histidine. Taken together, these solid-state NMR data show that inward rectification in M2 proton channels is accomplished by tryptophan interactions with charged residues on both its C-terminal and N-terminal sides. Gating of these M2 proton channels is thus accomplished by a multi-residue complex with finely tuned electrostatic and aromatic interactions.

DiffBindFR: an SE(3) equivariant network for flexible protein-ligand docking.

Molecular docking, a key technique in structure-based drug design, plays pivotal roles in protein-ligand interaction modeling, hit identification and optimization, in which accurate prediction of protein-ligand binding mode is essential. Conventional docking approaches perform well in redocking tasks with known protein binding pocket conformation in the complex state. However, in real-world docking scenario without knowing the protein binding conformation for a new ligand, accurately modeling the binding complex structure remains challenging as flexible docking is computationally expensive and inaccurate. Typical deep learning-based docking methods do not explicitly consider protein side chain conformations and fail to ensure the physical plausibility and detailed atomic interactions. In this study, we present DiffBindFR, a full-atom diffusion-based flexible docking model that operates over the product space of ligand overall movements and flexibility and pocket side chain torsion changes. We show that DiffBindFR has higher accuracy in producing native-like binding structures with physically plausible and detailed interactions than available docking methods. Furthermore, in the Apo and AlphaFold2 modeled structures, DiffBindFR demonstrates superior advantages in accurate ligand binding pose and protein binding conformation prediction, making it suitable for Apo and AlphaFold2 structure-based drug design. DiffBindFR provides a powerful flexible docking tool for modeling accurate protein-ligand binding structures.

Microwave irradiation: Effect of β-glucosidase activity by modulating its structural properties in grape maceration stage

In this paper, microwave as an auxiliary method could activate the β-glucosidase (β-GC) activity in grape maceration stage. β-GC had higher tolerance to acid and temperature during winemaking process. Microwave parameters were optimized for using response surface methodology (RSM), variables including microwave power, temperature, and time. The results demonstrated that the regression equation obtained by using the quadratic regression orthogonal model could better simulate actual operating conditions, the β-GC activity increased by 12.54% under optimum conditions (700 W, 48 °C, and 6.9 min) comparing to the untreated β-GC. The changes in β-GC structural properties confirmed that microwave irradiation did increase β-GC activity by affecting β-GC's side chain group, hydrolysis degree, microenvironment, stability, composition, and conformation. The study would contribute to understanding the mechanism of aroma enhancement by microwave maceration.

Plasticizer-Induced Enhancement of Mesoscale Dissymmetry in Thin Films of Chiral Polymers with Variable Chain Length.

Conjugated polymers with chiral side chains are of interest in areas including chiral photonics, optoelectronics, and chemical and biological sensing. However, the low dissymmetry factors of most neat polymer thin films have limited their practical application. Here, a robust method to increase the absorption dissymmetry factor in a poly-fluorene-thiophene (PF8TS series) system is demonstrated by varying molecular weight and introducing an achiral plasticizer, polyethylene mono alcohol (PEM-OH). Extending chain length within the optimal range and adding this long-chain alcohol significantly enhance the chiroptical properties of spin-coated and annealed thin films. Mueller matrix spectroscopic ellipsometry (MMSE) analysis shows good agreement with the steady-state transmission measurements confirming a strong chiral response (circular dichroism (CD)and circular birefringence (CB)), ruling out linear dichroism, birefringence, and specific reflection effects. Solid-state NMR studies of annealed hybrid chiral polymer systems show enhancement of signals associated with aromatic π-stacked backbone and the ordered side-chain conformations. Further studies using Raman spectroscopy, X-ray diffraction (XRD), differential scanning calorimetry (DSC), atomic force microscopy (AFM), and polarized optical microscopy (POM) indicate that PEM-OH facilitates mesoscopic crystal domain ordering upon annealing. This provides new insights into routes for tuning optical activity in conjugated polymers.

AlphaFold predictions are valuable hypotheses and accelerate but do not replace experimental structure determination

Artificial intelligence-based protein structure prediction methods such as AlphaFold have revolutionized structural biology. The accuracies of these predictions vary, however, and they do not take into account ligands, covalent modifications or other environmental factors. Here, we evaluate how well AlphaFold predictions can be expected to describe the structure of a protein by comparing predictions directly with experimental crystallographic maps. In many cases, AlphaFold predictions matched experimental maps remarkably closely. In other cases, even very high-confidence predictions differed from experimental maps on a global scale through distortion and domain orientation, and on a local scale in backbone and side-chain conformation. We suggest considering AlphaFold predictions as exceptionally useful hypotheses. We further suggest that it is important to consider the confidence in prediction when interpreting AlphaFold predictions and to carry out experimental structure determination to verify structural details, particularly those that involve interactions not included in the prediction.

Ab Initio Prediction of 3-D Conformations for Protein Long Loops with High Accuracy and Applications to Antibody CDRH3 Modeling.

Residue-level potentials of mean force were widely used for protein backbone refinements to avoid simultaneous sampling of side-chain conformations. The interaction energy between the reduced side chains and backbone atoms was not considered explicitly. In this study, we developed novel methods to calculate the residue-atom interaction energy in combination with atomic and residue-level terms. The parameters were optimized step by step to remove the overcounting or overlap problem between different energy terms. The mixing energy functions were then used to evaluate the generated backbone conformations at the initial sampling stage of protein loop modeling (OSCAR-loop), including the interaction energy between the reduced loop residues and full atoms of the protein framework. The accuracies of top-ranked decoys were 1.18 and 2.81 Å for 8-residue and 12-residue loops, respectively. We then selected diverse decoys for side-chain modeling, backbone refinement, and energy minimization. The procedure was repeated multiple times to select one prediction with the lowest energy. Consequently, we obtained an accuracy of 0.74 Å for a prevailing test set of 12-residue loops, compared with >1.4 Å reported by other researchers. The OSCAR-loop was also effective for modeling the H3 loops of antibody complementary determining regions (CDRs) in the crystal environment. The prediction accuracy of OSCAR-loop (1.74 Å) was better than the accuracy of the Rosetta NGK method (3.11 Å) or those achieved by deep learning methods (>2.2 Å) for the CDRH3 loops of 49 targets in the Rosetta antibody benchmark. The performance of OSCAR-loop in a model environment was also discussed.